tabalyst 0.1.0__tar.gz → 0.1.2__tar.gz

{tabalyst-0.1.0/src/tabalyst.egg-info → tabalyst-0.1.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: tabalyst
-Version: 0.1.0
+Version: 0.1.2
 Summary: Open-source data profiling engine for structured JSON and customizable HTML reports.
 Author: Gregory Borelli
 License-Expression: MIT
@@ -61,7 +61,7 @@ python -m pip install -e ".[dev]"
 Pass the CSV source and the complete HTML report path as positional arguments:
 
 ```console
-tabalyst data.csv reports/client-a.html
+tabalyst data.csv reports/report.html
 ```
 
 The report path must end in `.html`. Parent directories are created
@@ -69,8 +69,8 @@ automatically, and existing artifacts are replaced. A successful run creates:
 
 ```text
 reports/
-|-- client-a.html
-|-- client-a.json
+|-- report.html
+|-- report.json
 `-- executions.json
 ```
 
@@ -81,8 +81,8 @@ several named reports can coexist there.
 Common options are:
 
 ```console
-tabalyst data.csv reports/client-a.html --separator ";" --encoding cp1252
-tabalyst data.csv reports/client-a.html --config tabalyst.json
+tabalyst data.csv reports/report.html --separator ";" --encoding cp1252
+tabalyst data.csv reports/report.html --config tabalyst.json
 tabalyst --help
 tabalyst --version
 ```
@@ -96,7 +96,7 @@ import tabalyst
 
 result = tabalyst.analyze(
     "data.csv",
-    "reports/client-a.html",
+    "reports/report.html",
     separator=";",
     encoding="cp1252",
     config_path="tabalyst.json",
@@ -162,21 +162,20 @@ See the complete layout and maintenance notes in the
 - Physical and semantic type inference with explicit confidence and error rates.
 - Numeric, date, string-length, normalization, and value-distribution profiles.
 - A bounded raw-data preview while all records are analyzed.
-- Sortable and filterable HTML tables.
+- Sortable and filterable HTML tables in the self-contained Signature report.
 
 Raw strings are preserved. The complete CSV is currently loaded into memory.
-Report data remains local and Tabalyst adds no telemetry or remote analysis.
-Bootstrap, DataTables, and Google Fonts are loaded from pinned CDNs for the full
-interactive presentation; the Tabalyst template, theme, and report JavaScript are
-included in the Python package.
+Report data remains local and Tabalyst adds no telemetry or remote analysis. The
+report embeds its design system, font subsets and interaction code, so it works
+without CDN or network access.
 
 ## Compatibility with alpha commands
 
 The earlier commands remain available during the `0.1.x` transition:
 
 ```console
-tabalyst analyze data.csv -o reports/client-a.html
-tabalyst render reports/client-a.json -o reports/regenerated.html
+tabalyst analyze data.csv -o reports/report.html
+tabalyst render reports/report.json -o reports/regenerated.html
 ```
 
 The alpha `analyze` form also retains automatic `tabalyst.json` discovery,
@@ -194,6 +193,8 @@ python -m build
 The build creates a wheel and source distribution under `dist/`. Release steps,
 including the clean-wheel smoke test and PyPI Trusted Publishing setup, are in
 [RELEASING.md](https://github.com/loribel-labs/tabalyst/blob/main/RELEASING.md).
+The packaging architecture and automated publication flow are documented in
+[docs/python-package-and-release.md](https://github.com/loribel-labs/tabalyst/blob/main/docs/python-package-and-release.md).
 
 Report format details and internal boundaries are documented in
 [docs/architecture.md](https://github.com/loribel-labs/tabalyst/blob/main/docs/architecture.md).

{tabalyst-0.1.0 → tabalyst-0.1.2}/README.md

@@ -27,7 +27,7 @@ python -m pip install -e ".[dev]"
 Pass the CSV source and the complete HTML report path as positional arguments:
 
 ```console
-tabalyst data.csv reports/client-a.html
+tabalyst data.csv reports/report.html
 ```
 
 The report path must end in `.html`. Parent directories are created
@@ -35,8 +35,8 @@ automatically, and existing artifacts are replaced. A successful run creates:
 
 ```text
 reports/
-|-- client-a.html
-|-- client-a.json
+|-- report.html
+|-- report.json
 `-- executions.json
 ```
 
@@ -47,8 +47,8 @@ several named reports can coexist there.
 Common options are:
 
 ```console
-tabalyst data.csv reports/client-a.html --separator ";" --encoding cp1252
-tabalyst data.csv reports/client-a.html --config tabalyst.json
+tabalyst data.csv reports/report.html --separator ";" --encoding cp1252
+tabalyst data.csv reports/report.html --config tabalyst.json
 tabalyst --help
 tabalyst --version
 ```
@@ -62,7 +62,7 @@ import tabalyst
 
 result = tabalyst.analyze(
     "data.csv",
-    "reports/client-a.html",
+    "reports/report.html",
     separator=";",
     encoding="cp1252",
     config_path="tabalyst.json",
@@ -128,21 +128,20 @@ See the complete layout and maintenance notes in the
 - Physical and semantic type inference with explicit confidence and error rates.
 - Numeric, date, string-length, normalization, and value-distribution profiles.
 - A bounded raw-data preview while all records are analyzed.
-- Sortable and filterable HTML tables.
+- Sortable and filterable HTML tables in the self-contained Signature report.
 
 Raw strings are preserved. The complete CSV is currently loaded into memory.
-Report data remains local and Tabalyst adds no telemetry or remote analysis.
-Bootstrap, DataTables, and Google Fonts are loaded from pinned CDNs for the full
-interactive presentation; the Tabalyst template, theme, and report JavaScript are
-included in the Python package.
+Report data remains local and Tabalyst adds no telemetry or remote analysis. The
+report embeds its design system, font subsets and interaction code, so it works
+without CDN or network access.
 
 ## Compatibility with alpha commands
 
 The earlier commands remain available during the `0.1.x` transition:
 
 ```console
-tabalyst analyze data.csv -o reports/client-a.html
-tabalyst render reports/client-a.json -o reports/regenerated.html
+tabalyst analyze data.csv -o reports/report.html
+tabalyst render reports/report.json -o reports/regenerated.html
 ```
 
 The alpha `analyze` form also retains automatic `tabalyst.json` discovery,
@@ -160,6 +159,8 @@ python -m build
 The build creates a wheel and source distribution under `dist/`. Release steps,
 including the clean-wheel smoke test and PyPI Trusted Publishing setup, are in
 [RELEASING.md](https://github.com/loribel-labs/tabalyst/blob/main/RELEASING.md).
+The packaging architecture and automated publication flow are documented in
+[docs/python-package-and-release.md](https://github.com/loribel-labs/tabalyst/blob/main/docs/python-package-and-release.md).
 
 Report format details and internal boundaries are documented in
 [docs/architecture.md](https://github.com/loribel-labs/tabalyst/blob/main/docs/architecture.md).

{tabalyst-0.1.0 → tabalyst-0.1.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "tabalyst"
-version = "0.1.0"
+version = "0.1.2"
 description = "Open-source data profiling engine for structured JSON and customizable HTML reports."
 readme = "README.md"
 requires-python = ">=3.11"

{tabalyst-0.1.0 → tabalyst-0.1.2}/src/tabalyst/analysis.py

@@ -14,6 +14,8 @@ from tabalyst.config import AnalysisConfig
 from tabalyst.ingestion import CsvDataset
 from tabalyst.models import (
     ColumnProfile,
+    DatasetDateColumnSummary,
+    DatasetDateSummary,
     DatasetProfile,
     DatasetSummary,
     DateBreakdownItem,
@@ -313,10 +315,15 @@ def build_date_profile(
     return (
         DateProfile(
             status=status,
+            present_count=total,
             valid_count=valid_count,
+            valid_percent=percent(valid_count, total),
             ambiguous_count=ambiguous_count,
+            ambiguous_percent=percent(ambiguous_count, total),
             invalid_date_count=invalid_count,
+            invalid_date_percent=percent(invalid_count, total),
             not_date_count=not_date_count,
+            not_date_percent=percent(not_date_count, total),
             resolved_ambiguous_order=resolved_order,
             ambiguous_order_source=resolution_source,
             formats=format_items,
@@ -406,15 +413,19 @@ def build_string_profile(
         status = "very_long"
     distribution = []
     if maximum <= settings.length_distribution_max_length:
+        length_counts = lengths.value_counts()
+        maximum_length_count = int(length_counts.max())
         for length in sorted({int(value) for value in lengths}):
             matching = [
                 (value, count) for value, count in frequencies if len(value) == length
             ]
+            length_count = sum(count for _, count in matching)
             distribution.append(
                 StringLengthDistribution(
                     length=length,
-                    count=sum(count for _, count in matching),
-                    percent=percent(sum(count for _, count in matching), len(present)),
+                    count=length_count,
+                    percent=percent(length_count, len(present)),
+                    relative_percent=percent(length_count, maximum_length_count),
                     distinct_count=len(matching),
                     examples=[
                         StringLengthExample(value=value, count=count)
@@ -432,6 +443,11 @@ def build_string_profile(
         distinct_length_count=int(lengths.nunique()),
         fixed_length=minimum if minimum == maximum else None,
         length_distribution=distribution,
+        representative_examples=[
+            item.examples[0].value
+            for item in sorted(distribution, key=lambda item: -item.count)[:3]
+            if item.examples
+        ],
     )
 
 
@@ -632,7 +648,11 @@ def analyze_column(
         stats = [numbers.min(), numbers.max(), numbers.mean(), numbers.median()]
         if numbers.notna().all() and all(math.isfinite(value) for value in stats):
             numeric = NumericStats(
-                minimum=stats[0], maximum=stats[1], mean=stats[2], median=stats[3]
+                minimum=stats[0],
+                maximum=stats[1],
+                range=stats[1] - stats[0],
+                mean=stats[2],
+                median=stats[3],
             )
     value_profile = build_value_profile(
         frequencies,
@@ -649,6 +669,7 @@ def analyze_column(
     if enum and semantic_type is None:
         semantic_type = "enum"
     string_profile = build_string_profile(present, frequencies, inferred, config)
+    missing_count = int(absent.sum())
     return ColumnProfile(
         id=f"column_{position}",
         name=name if name is not None else str(values.name or ""),
@@ -658,8 +679,9 @@ def analyze_column(
         type_confidence=round(type_confidence, 4),
         type_error_count=type_error_count,
         type_error_percent=type_error_percent,
-        missing_count=int(absent.sum()),
-        missing_percent=percent(int(absent.sum()), len(values)),
+        missing_count=missing_count,
+        missing_percent=percent(missing_count, len(values)),
+        with_issues=missing_count > 0 or inferred == "mixed",
         normalization=normalization,
         distinct_count=len(frequencies),
         examples=[
@@ -691,6 +713,40 @@ def analyze_dataset(dataset: CsvDataset, config: AnalysisConfig) -> DatasetProfi
     empty = [column.id for column in columns if column.inferred_type == "empty"]
     constant = [column.id for column in columns if column.distinct_count == 1]
     mixed = [column.id for column in columns if column.inferred_type == "mixed"]
+    numeric_columns = [column for column in columns if column.numeric is not None]
+    date_columns = [column for column in columns if column.date_profile is not None]
+    string_columns = [column for column in columns if column.string_profile is not None]
+    inferred_type_counts = Counter(column.inferred_type for column in columns)
+    semantic_type_counts = Counter(column.semantic_type or "none" for column in columns)
+    date_present_count = sum(
+        column.date_profile.present_count for column in date_columns if column.date_profile
+    )
+    date_valid_count = sum(
+        column.date_profile.valid_count for column in date_columns if column.date_profile
+    )
+    date_ambiguous_count = sum(
+        column.date_profile.ambiguous_count
+        for column in date_columns
+        if column.date_profile
+    )
+    date_invalid_count = sum(
+        column.date_profile.invalid_date_count
+        for column in date_columns
+        if column.date_profile
+    )
+    date_not_date_count = sum(
+        column.date_profile.not_date_count
+        for column in date_columns
+        if column.date_profile
+    )
+    maximum_column_ambiguous_count = max(
+        (
+            column.date_profile.ambiguous_count
+            for column in date_columns
+            if column.date_profile
+        ),
+        default=0,
+    )
     issues = []
 
     def add_issue(
@@ -774,6 +830,47 @@ def analyze_dataset(dataset: CsvDataset, config: AnalysisConfig) -> DatasetProfi
             empty_row_count=int(absent.all(axis=1).sum()),
             empty_column_count=len(empty),
             constant_column_count=len(constant),
+            with_issues_column_count=sum(column.with_issues for column in columns),
+            numeric_column_count=len(numeric_columns),
+            date_column_count=len(date_columns),
+            string_column_count=len(string_columns),
+            inferred_type_counts=dict(inferred_type_counts),
+            inferred_type_percents={
+                name: percent(count, len(columns))
+                for name, count in inferred_type_counts.items()
+            },
+            semantic_type_counts=dict(semantic_type_counts),
+            semantic_type_percents={
+                name: percent(count, len(columns))
+                for name, count in semantic_type_counts.items()
+            },
+        ),
+        date_summary=(
+            DatasetDateSummary(
+                present_count=date_present_count,
+                valid_count=date_valid_count,
+                ambiguous_count=date_ambiguous_count,
+                ambiguous_percent=percent(date_ambiguous_count, date_present_count),
+                invalid_date_count=date_invalid_count,
+                not_date_count=date_not_date_count,
+                maximum_column_ambiguous_count=maximum_column_ambiguous_count,
+                columns=[
+                    DatasetDateColumnSummary(
+                        column_id=column.id,
+                        name=column.name,
+                        position=column.position,
+                        ambiguous_count=column.date_profile.ambiguous_count,
+                        relative_ambiguous_percent=percent(
+                            column.date_profile.ambiguous_count,
+                            maximum_column_ambiguous_count,
+                        ),
+                    )
+                    for column in date_columns
+                    if column.date_profile and column.date_profile.ambiguous_count
+                ],
+            )
+            if date_columns
+            else None
         ),
         columns=columns,
         issues=issues,

{tabalyst-0.1.0 → tabalyst-0.1.2}/src/tabalyst/models.py

@@ -23,6 +23,7 @@ class SourceInfo(ResultModel):
 class NumericStats(ResultModel):
     minimum: FiniteFloat
     maximum: FiniteFloat
+    range: FiniteFloat
     mean: FiniteFloat
     median: FiniteFloat
 
@@ -52,10 +53,15 @@ class DateBreakdownItem(ResultModel):
 
 class DateProfile(ResultModel):
     status: Literal["valid", "multiple_formats", "mixed", "ambiguous", "invalid"]
+    present_count: int
     valid_count: int
+    valid_percent: float
     ambiguous_count: int
+    ambiguous_percent: float
     invalid_date_count: int
+    invalid_date_percent: float
     not_date_count: int
+    not_date_percent: float
     resolved_ambiguous_order: Literal["MDY", "DMY"] | None = None
     ambiguous_order_source: Literal["config", "column"] | None = None
     formats: list[DateFormatCount]
@@ -73,6 +79,7 @@ class StringLengthDistribution(ResultModel):
     length: int
     count: int
     percent: float
+    relative_percent: float
     distinct_count: int
     examples: list[StringLengthExample]
 
@@ -87,6 +94,7 @@ class StringProfile(ResultModel):
     distinct_length_count: int
     fixed_length: int | None = None
     length_distribution: list[StringLengthDistribution]
+    representative_examples: list[str]
 
 
 class ValueOccurrence(ResultModel):
@@ -122,6 +130,7 @@ class ColumnProfile(ResultModel):
     type_error_percent: float | None
     missing_count: int
     missing_percent: float
+    with_issues: bool
     normalization: NormalizationStats
     distinct_count: int
     examples: list[str]
@@ -145,6 +154,33 @@ class DatasetSummary(ResultModel):
     empty_row_count: int
     empty_column_count: int
     constant_column_count: int
+    with_issues_column_count: int
+    numeric_column_count: int
+    date_column_count: int
+    string_column_count: int
+    inferred_type_counts: dict[str, int]
+    inferred_type_percents: dict[str, float]
+    semantic_type_counts: dict[str, int]
+    semantic_type_percents: dict[str, float]
+
+
+class DatasetDateColumnSummary(ResultModel):
+    column_id: str
+    name: str
+    position: int
+    ambiguous_count: int
+    relative_ambiguous_percent: float
+
+
+class DatasetDateSummary(ResultModel):
+    present_count: int
+    valid_count: int
+    ambiguous_count: int
+    ambiguous_percent: float
+    invalid_date_count: int
+    not_date_count: int
+    maximum_column_ambiguous_count: int
+    columns: list[DatasetDateColumnSummary]
 
 
 class Issue(ResultModel):
@@ -163,12 +199,13 @@ class PreviewRow(ResultModel):
 
 class DatasetProfile(ResultModel):
     format_version: Literal["0.1.0a"] = "0.1.0a"
-    format_revision: Literal[1] = 1
+    format_revision: Literal[2] = 2
     generated_at: datetime
     processing_seconds: FiniteFloat
     source: SourceInfo
     config: AnalysisConfig
     summary: DatasetSummary
+    date_summary: DatasetDateSummary | None = None
     columns: list[ColumnProfile]
     issues: list[Issue]
     preview: list[PreviewRow]

{tabalyst-0.1.0 → tabalyst-0.1.2}/src/tabalyst/reporting.py

@@ -1,6 +1,5 @@
 """Render self-contained analysis results as an interactive HTML report."""
 
-from collections import Counter
 from importlib.resources import files
 from pathlib import Path
 
@@ -45,10 +44,30 @@ def render_report(profile: DatasetProfile) -> str:
     template = environment.from_string(
         resources.joinpath("templates/report.html").read_text(encoding="utf-8")
     )
+    type_colors = {
+        "text": "var(--d2)",
+        "mixed": "var(--attn)",
+        "integer": "var(--d3)",
+        "number": "var(--d1)",
+        "boolean": "var(--ok)",
+        "date": "var(--d1)",
+        "empty": "var(--line-strong)",
+    }
+    semantic_colors = {
+        "enum": "var(--d2)",
+        "date": "var(--d1)",
+        "none": "var(--d3)",
+    }
     return template.render(
         report=profile,
         tabalyst_version=get_version(),
-        types=Counter(column.inferred_type for column in profile.columns),
+        source_stem=Path(profile.source.filename).stem,
+        source_suffix=Path(profile.source.filename).suffix,
+        numeric_columns=[column for column in profile.columns if column.numeric],
+        date_columns=[column for column in profile.columns if column.date_profile],
+        string_columns=[column for column in profile.columns if column.string_profile],
+        type_colors=type_colors,
+        semantic_colors=semantic_colors,
         theme_css=resources.joinpath("static/theme.css").read_text(encoding="utf-8"),
         report_js=resources.joinpath("static/report.js").read_text(encoding="utf-8"),
     )